Interdisziplinäres Zentrum
für Bioinformatik
|
|
Interdisziplinäres Zentrum |
|
| Home | Suche | Sitemap | Impressum | |  |
  
|
Sequence-specific hybridization of GeneChip microarraysHans Binder, Stephan Preibisch Interdisciplinary Centre for Bioinformatics University of Leipzig Cooperation: Markus Löffler, Ernst Schuster, Dirk Hasenclever, Ingo Röder Institute for Medical Informatics, Statistics and Epidemiology University of Leipzig Friedemann Horn  Institute of Clinical Immunology and Transfusion Medicine University Hospital Leipzig Jürgen Läuter, Maciej Rosolowski, Toralf Kirsten, Jörg Lange IZBI University of Leipzig Peter F. Stadler Dept. of Computer Science University of Leipzig 
The “riddle of bright MM”: The correlation between the intensities of perfect matched (PM) and mismatched (MM)
probes of an Affymetrix microarray reveals that a fair number of the MM shine brighter than the PM, i.e. one observes PM>MM and PM<MM as well.
This puzzling property refers however only to non-specifically hybridized probes (blue dots). For specifically hybridized probes one observes
nearly exclusively PM>MM (red dots). In reality, this straightforward working principle is opposed by the complexity of the experimental system caused by imperfections of chip fabrication and RNA preparation, and especially by several competitive interactions which are inherently connected with the high throughput character of the method. The raw intensity data are affected by these parasitic interferences which prohibit the direct translation of the probe intensities into the “expression degree” as a measure of the concentration of the target RNA referring to the genomic sequence of interest. The project addresses this issue in terms of a hybridization model to correct raw microarray data for the effect of (i) Sequence-specific affinities; (ii) mismatches; (iii) cross-hybridization and (iv) saturation. These tasks are essential prerequisites for developing adequate analysis algorithms of microarray data. Results: We derived the basic model of microarray hybridization and systematically analyzed the specific effect of the probe sequence on the observed intensities 1-3. It was for example shown that base-specific contributions to the intensity are the dominating term of the signal variability and, moreover, they well correlate with base-specific affinity parameter for DNA/RNA hybridization in solution. The second aspect of the project adressed molecular details of probe-target interactions to refine the model 4-6.Particularly, the so-called “riddle of bright MM” was solved, a fundamental problem which hampered the adequate inclusion of mismatched probes into the analysis of Affymetrix chips. As a second important result we found a molecular signature of specific and non-specific hybridization. The next step of the project dealt with the response function of the microarrays to changes of the RNA-concentrations and with the choice of appropriate corrections to extract the expression degree 7, 8. Different effects which affect the hybridization of DNA-oligonucleotide probes such as the interactions between nucleotide strands competing with specific duplex formation in a complex multicomponent mixture of RNA fragments, electrostatic and entropic blocking, the fragmentation of the RNA, the incomplete synthesis of the probes and “zippering” effects were analyzed. 
Solving the riddle of bright MM: The intensity relation between the PM and MM probes can be rationalized in
terms of a molecular signature of the central base pairing formed by the middle base of the DNA-probe (capital letter) with specific
(i.e. completely complementary) and non-specific (i.e. partly complementary) RNA (lower case letter). The PM form exclusively Watson-Crick
pairings (C•g in the example). Specifically hybridized MM possess however a self-complementary central pairing (G•g). The stability of
Watson-Crick pairings is always stronger than that of self-complementary pairings giving rise to “bright” PM in the limit of specific
hybridization, i.e. PM>MM. For non-specifically hybridized MM the Watson-Crick pairing reverses direction (G•c). Watson-Crick pairings
in DNA/RNA hybrid duplexes are asymmetric with respect to the reversal of the bond: A purine (e.g., C or T) in the DNA binds stronger
than the respective pyrimidine (G or A) giving rise either to PM>MM or PM<MM for the intensities of non-specifically hybridized duplexes.
The bases A and T show analogous properties as G and C.
Summary and Outlook: Our hybridization model explicitly considers the RNA concentration and the amount of specific and non-specific transcripts in the sample solution on the one hand and the sequence of the oligonucleotide probes on the other hand. It explains the concentration dependence and the effect of the middle base on the intensity of perfect matched and mismatched microarray probes. The actual work focuses on the transfer of the obtained results into sophisticated analysis algorithms of Affymetrix microarray expression data. Its advantage suggests future applications to genomic SNP- and tiling arrays, and special analyses such as the search for alternative splicing variants and cross-species hybridizations.
 top
|